Coordinating Paleoclimate Research on Tropical Cyclones with Hurricane-Climate Theory and Modelling
Total Page:16
File Type:pdf, Size:1020Kb
Tellus (2007), 59A, 529–537 C 2007 The Authors Journal compilation C 2007 Blackwell Munksgaard Printed in Singapore. All rights reserved TELLUS Perspective: coordinating paleoclimate research on tropical cyclones with hurricane-climate theory and modelling ∗ By AMY FRAPPIER1 , THOMAS KNUTSON2, KAM-BIU LIU3 and KERRY EMANUEL4 1Department of Geology and Geophysics, Boston College, Devlin Hall 213, Chestnut Hill, MA 02467 USA 2Geophysical Fluid Dynamics Laboratory/NOAA, Princeton, NJ 08542 USA 3Department of Oceanography and Coastal Sciences, Louisiana State University, 1002 Energy, Coast, and Environment Building, Baton Rouge, LA 70803, USA 4Program in Atmospheres, Oceans, and Climate, MIT, Cambridge, MA 02139 USA (Manuscript received 1 September 2006; in final form 3 April 2007) ABSTRACT Extending the meteorological record back in time can offer critical data for assessing tropical cyclone-climate links. While paleotempestology, the study of ancient storms, can provide a more realistic view of past ‘worst case scenarios’, future environmental conditions may have no analogues in the paleoclimate record. The primary value in paleotem- pestology proxy records arises from their ability to quantify climate–tropical cyclone interactions by sampling tropical cyclone activity during pre-historic periods with a wider range of different climates. New paleotempestology proxies are just beginning to be applied, encouraging new collaboration between the paleo and tropical cyclone dynamics com- munities. The aim of this paper is to point out some paths toward closer coordination by outlining target needs of the tropical cyclone theory and modelling community and potential contributions of the paleotempestology community. We review recent advances in paleotempestology, summarize the range of types and quality of paleodata generation, and identify future research opportunities for paleotempestology, tropical cyclone dynamics and climate change impacts and attribution communities. orological record back in time by exploiting documentary and 1. Introduction geological/biological proxy archives of TC activity (Liu, 2004; Recent advances in the field of tropical cylone (TC) dynam- Nott, 2004; Liu, 2007). ics have raised new questions about the importance of anthro- Paleotempestology records provide a better estimate of the pogenic climate change for future hurricane hazards in the near ‘worst case scenario’ than conventional historical hurricane and long term (Elsner and Liu, 2003; Emanuel, 2001, 2002, databases because very long records are more likely to sam- 2003; Camargo and Sobel, 2004; De et al., 2004; Elsner et al., ple very rare, catastrophic events with long recurrence inter- 2004; Emanuel and Nolan, 2004; Emanuel et al., 2004; Free vals of hundreds to thousands of years (Liu and Fearn, 1993; et al., 2004; Murnane, 2004; Walsh, 2004; Wu and Wang, 2004; Liu and Fearn, 2000; Nott and Hayne, 2001; Nott, 2003a). For Emanuel, 2005; Pielke et al., 2005; Trenberth, 2005; Webster example, Liu (2004) documented a Gulf Coast ‘hyperactive et al., 2005). Reanalysis and documentary-based TC reconstruc- period’ about 3800 to 1000 yr ago during which catastrophic tions can overcome some of the limitations inherent in histor- hurricanes struck 3 to 5 times more frequently than during the ical meteorological data sets of TC activity (Liu et al., 2001; most recent millennium. Because, the ‘best case’ scenario—no Chenoweth, 2003; Landsea et al., 2004; Garc´ıa-Herrera et al., major landfalls in populated areas—plays itself out regularly, 2005; Landsea et al., 2006; Kossin et al., 2007; Mock, 2004). paleotempestology is often the only source of data regarding Yet, regional gaps and the brevity of meteorological records re- very large/rare events in the tail of the distribution. While ‘hy- main problematic. Paleotempestology aims to extend the mete- peractive’ and ‘inactive’ periods are of similar interest for study- ing TC dynamics, the climatology of extreme catastrophic cases ∗ is of far greater interest to society, which must bear the costs Corresponding author. e-mail: [email protected] mainly when TCs strike occupied locales. Brief meteorological No claim to U.S. government works. data sets sometimes grossly undersample the landfall probability DOI: 10.1111/j.1600-0870.2007.00250.x of catastrophic TCs (Liu, 2004, 2007; Nott, 2003a; Nott et al., Tellus 59A (2007), 4 529 530 A. FRAPPIER ET AL. 2007). Still, the concept of ‘recurrence interval’ assumes an in- For example, probability theory could be applied to estimate the variant probability field. To the extent that TCs are a manifesta- likelihood of a given number or fraction of independent sites tion of shifting climatic conditions, the spectrum of recurrence showing changes of a given sign in global TC frequency, result- interval risk is also non-stationary. Yet, expected future climate ing in adaptive estimates of sampling requirements needed to conditions forced by anthropogenic greenhouse gases lack close obtain certain confidence levels for changes in global TC fre- paleoclimatic analogues (e.g. Crowley, 1990; Mitchell, 1990; quency. Scenarios in which inferred TC changes relatively large Covey et al., 1996;). For example, the Holocene Climatic Opti- might be addressed sooner, as fewer sites would be sufficient mum (∼6000–9000 y.b.p.), a warm period when many climatic to determine first-order sensitivity as a function of past climate boundary conditions were similar to today, was characterized by conditions. seasonal rather than mean annual warming (Crowley, 1990). Pa- leotempestology data alone are unlikely to provide an accurate 2.2. Sampling of tropical cyclone behaviour from very estimate of the future probability of extreme events. On the other different climates hand, paleotempestology records that have been produced from more recent intervals provide better analogues to present and Each proxy type has a different applicable time frame (Table 1). future climates. Thus, the most valuable contribution of pale- Very long TC records (e.g. thousands of years) provide better otempestology will likely be as a unique source of observations sampling of rare extreme events. If we assume that the TC cli- critical for assessing tropical cyclone–climate interactions across mate is stationary during the period, this allows investigators to a range of climatic conditions (Cronin, 1999). better estimate the underlying (stationary) probability density A growing suite of high-resolution paleotempestology data function. This information, useful for documenting the amount has the potential to shed light on problems with which the of past variability in the real world, is necessary for assessing broader TC dynamics community is wrestling (Cohen, 2001; the degree to which any secular trend (e.g. warming during the Miller, 2005; Miller et al., 2006; Frappier et al, 2007; Nott, instrumental era) is unusual with respect to past variations, as has 2007). The aim of this paper is to point out some ways toward been done by several groups for the case of Northern Hemisphere closer coordination between the TC dynamics and paleotem- mean temperatures over the past millennium (e.g. Mann et al., pestology communities by outlining target needs of the former 1999; Jones et al., 2001; Mann et al., 2003; Jones and Mann, and potential contributions of the latter. We review recent ad- 2004; Rutherford et al., 2005). Analogously, climate change de- vances in paleotempestology, summarize the potential types and tection and attribution studies for TCs stand to benefit from fur- quality of paleodata generation and identify opportunities for ther development and analysis of paleotempestology data sets. collaboration. Paleotempestology data can also define rates of change in storm activity parameters (e.g. Liu et al., 2001; Elsner et al, 2004; Zhao and Chu, 2006). 2. Tropical cyclone theory and modelling Paleoclimatic variability provides a series of natural exper- efforts: potential contributions from iments from which valuable observational constraints regard- paleotempestology ing TC-climate sensitivity can be derived by sampling TC be- haviour during very different climates. Independent proxy re- 2.1. Substantial regional coverage of long-time records constructions of paleo-TCs, paleoclimate boundary conditions of tropical cyclone activity and forcing factors can provide well-grounded estimates of cli- Attempts to reconstruct paleo-temperature from various proxy mate change and therefore climate sensitivity, assuming the measures at a limited number of locations, analogous to mete- climate forcing can be estimated well enough. For TCs, the orological stations (e.g. Mann et al. 1999), are made more fea- Last Glacial Maximum may provide a sufficiently large climate sible by the relatively large spatial correlation scales of long change signal that a robust TC response can be detected in proxy timescale temperature anomalies. For global paleotempestology records. With sufficient regional coverage, this finding could reconstructions, spatial correlations must also be considered, serve as a key evaluation point for models that are being used as modern TC activity increases in one region are sometimes to simulate TC-climate connections with respect to global-scale offset by decreases in another basin (c.f., Atlantic vs. Eastern changes in climate (Hamilton and Hemler, 1997; Henderson- Pacific). Therefore, obtaining a reliable picture of past global Sellers et al.,1998; Knutson et al, 1998; Knutson et al., 2001; TC occurrence might require